Easing Core Measure Pain, Overcoming 'Big Data'

The goal of "core measures," is not to measure. Rather, it's to drive improved performance across health care organizations and depending on whom you ask, it's to police hospitals to ensure certain patients are treated in line with what has been determined to be standard and necessary based on their diagnosis.

Core measures were originally spun out of the Joint Commission nearly 15 years ago as part of hospital accreditation. And today, they are tied to the Centers for Medicare & Medicaid Services' (CMS) value-based purchasing program, to emerging pay-for-performance initiatives, to the National Quality Forum, and are even made available online to help patients choose where they go for care. Increasingly, the pressure to comply with core measures is weighing heavily on health care organizations and clinical staff -- as noted above, their reputation, reimbursement and essential livelihood depends on them.

Top of Mind

Core measures are one of those things that are simply always on the minds of clinical staff. It's not so much because individual hospital workers have taken it upon themselves to drive compliance, but in large part it's because hospitals have set up mechanisms to actually remind caregivers of the need for core measure excellence. In many instances these reminders are thought to be disruptive. They are part of broader resource-intensive and non-real-time core measure management plans that may get the job (reporting) done, but in many cases are starved for improvement.

Not only are there built-in reminders as part of many health care IT platforms, but hospitals devote teams (documentation staff, nurses, etc.) to ensure core measure compliant care is provided when necessary and that the proper information is collected and later reported upon. This work is extremely manually intensive and oftentimes happens too slow for concurrent action to take place that could positively impact patient care. The reality of the situation is that while core measures are not going away anytime soon, health care organizations need to start thinking about how they can leverage automated technology to make workflow more efficient and effective, extend the value of staff (clinical and administrative), reduce overall costs, and most importantly as a means to ensure patients remain the focal point.

Core Measures are Complex

Currently, there are 58 general medical and surgical core measures, for which 179 distinct data elements are used to define. For example, acute myocardial infarction (AMI) is defined by a combination of what are called inclusion, exclusion and outcome data elements, including facts like: patient birth date, if the patient was or was not involved in a clinical trial, was a beta blocker administered, etc. For all core measures, a variety of pieces of information must be pulled from a number of sources: emergency department documentation, physician documentation, nursing documentation, the discharge summary, and more. The ongoing challenge for clinical staff is in finding data that in most cases is locked inside unstructured documentation. Some people are quick to advocate that health care organizations should replace unstructured documentation (text blobs in many cases) with purely structured data; this is risky business. In many cases unstructured documentation is built from the physician narrative -- it accounts for approximately 80 percent of all clinical information today and is usually built from highly-detailed notes about patients' unique health stories. For many caregivers, being able to refer to these notes from physician peers and from their own past patient encounters is critical as part of the patient recall process and for ongoing delivery of high levels of patient care (fully informed clinical decisions). The ideal documentation scenario is when both the patient story is preserved and structured data is captured, as this ensures both comprehensive data and streamlined abstraction capabilities.

Goodbye Retrospective, Hello Concurrency

By leveraging "understanding" and analytics technology, the health care industry can eradicate the manual review of clinical data. Not only is it time consuming and in many cases taking caregivers' attention away from patients, but it's ineffective. Simply put, in healthcare, retrospective performance reporting doesn't work. It doesn't enable real-time action adjustment or attention to a patient in need. Because of this, concurrency must become the norm -- the ability to examine what's happening across the care spectrum, while the patient is in the hospital, so proper care can be delivered and reporting is accurate and complete.

Through analytics automation health care organizations can identify key clinical data elements, which in many cases translate into patients in need, even if they live within an unstructured data source. By doing so, hospitals will be positioned well to manage the pressures of core measures and to increase the amount of clinical information they have real access to -- no longer will analysis depend on the work and attention of a skilled nurse.

What Becomes Possible

There are a number of data queries that are often run across hospitals to identify things like patients at risk for hospital-acquired conditions, present on admission indicators, patient safety indicators, and let's not forget core measure qualifiers. Today, as mentioned above, these queries are often done manually, but this doesn't need to be the case. With "understanding" and analytics technology health care organizations can do a variety of common and even custom searches against data. For example, within seconds an ad hoc query on "patients with heart failure" can be run to identify all patients with heart failure, past and present. Today there is analytics technology that, for lack of a better descriptor is actually smart enough to exclude patients who simply have "heart failure" as part of the family history note; only patients with heart failure will be presented. Beyond this, health care organizations can drill down into the data returned to identify the sickest of all patients, by searching for something like "troponin" or the ejection fraction percentage of heart failure patients. The queries that a hospital can perform when leveraging purpose-built "understanding" and analytics technology for health care is near endless: search for patients who've taken "Paxil" and "Pravachol" to identify high blood glucose levels, or quickly find patients with "sepsis" by identifying patient records where relevant terms like "septic shock," "septic," and "septicemia" exist or relevant clinical indicators exist like increased temp, heart rate, and respirations to ensure these patients are treated in line with the sepsis protocol.

On the Brink of Something Pretty Remarkable

There's a tremendous amount of valuable information in health care, but also a "big data" Rubik's cube problem. The obstacle is how to tap into that data to make sense of what's going on. Because we're faced with a mixture of structured and unstructured information, and a growing set of regulations to comply with, data and clinical language understanding technologies will increasingly become more prevalent and ubiquitous -- which to that we should all be relieved. We're facing a perfect storm in healthcare where industry need and technology robustness is finally aligning. From a patient's perspective, analytics and big data will aid, for example, in determining which hospital is the best for treating their condition. For healthcare organizations, data will become a tool, not a hindrance, and finally a foundation for true decision-support will be established. We're on the brink of better access to information - cracking the code, so to speak, of the big data problem in healthcare.